Configuring an AI node is where your workflow transitions from a structural plan to an active process. It is the point of execution. In this lesson, we examine the mechanics of setting up a node: selecting the appropriate model, defining its behavior through a system prompt, and providing the necessary input. Whether you are using OpenAI for complex reasoning or a faster model like Kimi for quick summarization, the configuration process remains consistent.
Beyond simply getting a response, we will focus on understanding the output. The lab provides detailed metricsâtoken usage, latency, and cost estimatesâthat are critical for building efficient workflows. We will also explore the A/B comparison panel, a tool designed to help you empirically determine which model is best suited for your specific task by testing them side by side.
Assignment
Create a new AI node in your workspace. Select a model of your choice and write a system prompt instructing it to summarize text. Provide a short paragraph as user input and run the node. Then, use the A/B comparison panel to test the same prompt against a different model. Note the differences in response quality, token usage, and latency.
Learning Objectives
- Understand how to select the appropriate AI model for a specific task.
- Learn to write effective system prompts and provide user input.
Model Selection
Choosing the right model (e.g., Qwen, DeepSeek, Kimi, Claude, OpenAI) based on the task requirements, balancing capability, speed, and cost.
Prompt Configuration
The process of defining a system prompt to set the AI's behavior and providing user input, which can be manual or chained from a previous node's output.